[1] We present a global chemical transport model called the Integrated Massively Parallel Atmospheric Chemical Transport (IMPACT) model. This model treats chemical and physical processes in the troposphere, the stratosphere, and the climatically critical tropopause region, allowing for physically based simulations of past, present, and future ozone and its precursors. The model is driven by meteorological fields from general circulation models (GCMs) or assimilated fields representing particular time periods. It includes anthropogenic and natural emissions, advective and convective transport, vertical diffusion, dry deposition, wet scavenging, and photochemistry. Simulations presented here use meteorological fields from the National Center for Atmospheric Research (NCAR) Middle Atmospheric Community Climate Model, Version 3 (MACCM3). IMPACT simulations of radon/lead are compared to observed vertical profiles and seasonal cycles. IMPACT results for a full chemistry simulation, with approximately 100 chemical species and 300 reactions representative of a mid-1990s atmosphere, are presented. The results are compared with surface, satellite, and ozonesonde observations. The model calculates a total annual flux from the stratosphere of 663 Tg O 3 /year, and a net in situ tropospheric photochemical source (that is, production minus loss) of 161 Tg O 3 /year, with 826 Tg O 3 /year dry deposited. NO x is overpredicted in the lower midlatitude stratosphere, perhaps because model aerosol surface densities are lower than actual values or the NO x to NO y conversion rate is underpredicted. Analysis of the free radical budget shows that ozone and NO y abundances are simulated satisfactorily, as are HO x catalytic cycles and total production and removal rates for ozone.

We investigate the sensitivity of precipitation characteristics (mean, extreme, and diurnal cycle) to a set of uncertain parameters that influence the qualitative and quantitative behavior of cloud and aerosol processes in the Community Atmosphere Model (CAM5). We adopt both the Latin hypercube and QuasiMonte Carlo sampling approaches to effectively explore the high-dimensional parameter space and then conduct two large sets of simulations. One set consists of 1100 simulations (cloud ensemble) perturbing 22 parameters related to cloud physics and convection, and the other set consists of 256 simulations (aerosol ensemble) focusing on 16 parameters related to aerosols and cloud microphysics. In the cloud ensemble, six parameters having the greatest influences on the global mean precipitation are identified, three of which (related to the deep convection scheme) are the primary contributors to the total variance of the phase and amplitude of the precipitation diurnal cycle over land. The extreme precipitation characteristics are sensitive to a fewer number of parameters. Precipitation does not always respond monotonically to parameter change. The influence of individual parameters does not depend on the sampling approaches or concomitant parameters selected. Generally, the Generalized Linear Model is able to explain more of the parametric sensitivity of global precipitation than local or regional features. The total explained variance for precipitation is primarily due to contributions from the individual parameters (75-90% in total). The total variance shows a significant seasonal variability in midlatitude continental regions, but very small in tropical continental regions.

No abstract

We systematically explore the ability of the Community Atmospheric Model version 5 (CAM5) to simulate the Madden-Julian Oscillation (MJO), through an analysis of MJO metrics calculated from a 1100-member perturbed parameter ensemble of 5 year simulations with observed sea surface temperatures. Parameters from the deep convection scheme make the greatest contribution to the variance in MJO simulation quality with a much smaller contribution from parameters in the large-scale cloud, shallow convection, and boundary layer turbulence schemes. Improved MJO variability results from a larger lateral entrainment rate and a reduction in the precipitation efficiency of deep convection that was achieved by a smaller autoconversion of cloud to rainwater and a larger evaporation of convective precipitation. Unfortunately, simulations with an improved MJO also have a significant negative impact on the climatological values of low-level cloud and absorbed shortwave radiation, suggesting that structural in addition to parametric modifications to CAM5's parameterization suite are needed in order to simultaneously well simulate the MJO and mean-state climate.

[1] Modern climate models contain numerous input parameters, each with a range of possible values. Since the volume of parameter space increases exponentially with the number of parameters N, it is generally impossible to directly evaluate a model throughout this space even if just 2-3 values are chosen for each parameter. Sensitivity screening algorithms, however, can identify input parameters having relatively little effect on a variety of output fields, either individually or in nonlinear combination. This can aid both model development and the uncertainty quantification (UQ) process. Here we report results from a parameter sensitivity screening algorithm hitherto untested in climate modeling, the Morris one-at-a-time (MOAT) method. This algorithm drastically reduces the computational cost of estimating sensitivities in a high dimensional parameter space because the sample size grows linearly rather than exponentially with N. It nevertheless samples over much of the N-dimensional volume and allows assessment of parameter interactions, unlike traditional elementary one-at-a-time (EOAT) parameter variation. We applied both EOAT and MOAT to the Community Atmosphere Model (CAM), assessing CAM's behavior as a function of 27 uncertain input parameters related to the boundary layer, clouds, and other subgrid scale processes. For radiation balance at the top of the atmosphere, EOAT and MOAT rank most input parameters similarly, but MOAT identifies a sensitivity that EOAT underplays for two convection parameters that operate nonlinearly in the model. MOAT's ranking of input parameters is robust to modest algorithmic variations, and it is qualitatively consistent with model development experience.

[1] Representation of clouds remains among the largest uncertainties in climate models and thus climate projections. Clouds vary significantly over different climate regimes and are controlled by different dynamics and physics. Using the cloud simulator output from perturbed-parameter ensemble climate runs with prescribed monthly sea surface temperature, this study examines the performance of the Community Atmosphere Model version 4 (CAM4) in simulating clouds over different tropical regions. Perturbing 28 selected parameters shows that model performance is quite sensitive to parameter values in different cloud regimes. Carefully adjusting these parameters could lead to a better simulation of clouds over many regions compared with the default model. Latin hypercube runs that pseudo-randomly sample the 28 parameters simultaneously have much wider spread and more spatial variations than the runs with parameters varied One-At-a-Time (OAT), suggesting the importance of non-linearities and interactions among parameters associated with different physical processes. The perturbed parameters have a relatively large impact on the mean bias compared to the pattern error. Citation:

Abstract. Simulations using IPCC (Intergovernmental Panel on Climate Change)-class climate models are subject to fail or crash for a variety of reasons. Quantitative analysis of the failures can yield useful insights to better understand and improve the models. During the course of uncertainty quantification (UQ) ensemble simulations to assess the effects of ocean model parameter uncertainties on climate simulations, we experienced a series of simulation crashes within the Parallel Ocean Program (POP2) component of the Community Climate System Model (CCSM4). About 8.5 % of our CCSM4 simulations failed for numerical reasons at combinations of POP2 parameter values. We applied support vector machine (SVM) classification from machine learning to quantify and predict the probability of failure as a function of the values of 18 POP2 parameters. A committee of SVM classifiers readily predicted model failures in an independent validation ensemble, as assessed by the area under the receiver operating characteristic (ROC) curve metric (AUC > 0.96). The causes of the simulation failures were determined through a global sensitivity analysis. Combinations of 8 parameters related to ocean mixing and viscosity from three different POP2 parameterizations were the major sources of the failures. This information can be used to improve POP2 and CCSM4 by incorporating correlations across the relevant parameters. Our method can also be used to quantify, predict, and understand simulation crashes in other complex geoscientific models.

Simulations using IPCC-class climate models are subject to fail or crash for a variety of reasons. Quantitative analysis of the failures can yield useful insights to better understand and improve the models. During the course of uncertainty quantification (UQ) ensemble simulations to assess the effects of ocean model parameter uncertainties on climate simulations, we experienced a series of simulation crashes within the Parallel Ocean Program (POP2) component of the Community Climate System Model (CCSM4). About 8.5% of our CCSM4 simulations failed for numerical reasons at combinations of POP2 parameter values. We apply support vector machine (SVM) classification from machine learning to quantify and predict the probability of failure as a function of the values of 18 POP2 parameters. A committee of SVM classifiers readily predicts model failures in an independent validation ensemble, as assessed by the area under the receiver operating characteristic (ROC) curve metric (AUC > 0.96). The causes of the simulation failures are determined through a global sensitivity analysis. Combinations of 8 parameters related to ocean mixing and viscosity from three different POP2 parameterizations are the major sources of the failures. This information can be used to improve POP2 and CCSM4 by incorporating correlations across the relevant parameters. Our method can also be used to quantify, predict, and understand simulation crashes in other complex geoscientific models